Journal ofMaterials Chemistry A

PAPER

Publ

ishe

d on

18

Sept

embe

r 20

15. D

ownl

oade

d by

Upp

sala

Uni

vers

ity o

n 09

/12/

2015

12:

40:1

1.

View Article OnlineView Journal | View Issue

Chemical engine

aUppsala University, Department of Ch

Chemistry, Box 523, SE 751 20 Uppsala, SwbUppsala University, Department of Physics a

Matter Physics, Box 516, SE 751 20 UppsalacUppsala University, Department of Che

Chemistry, Box 538, SE 751 21 Uppsala, Sw

† Electronic supplementary informa10.1039/c5ta05470b

Cite this: J. Mater. Chem. A, 2015, 3,21760

Received 17th July 2015Accepted 18th September 2015

DOI: 10.1039/c5ta05470b

www.rsc.org/MaterialsA

21760 | J. Mater. Chem. A, 2015, 3, 217

ering of methylammonium leadiodide/bromide perovskites: tuning of opto-electronic properties and photovoltaicperformance†

Byung-wook Park,a Bertrand Philippe,b Sagar M. Jain,a Xiaoliang Zhang,a

Tomas Edvinsson,c Hakan Rensmo,b Burkhard Zietza and Gerrit Boschloo*a

Hybrid (organic–inorganic) lead trihalide perovskites have attracted much attention in recent years due to

their exceptionally promising potential for application in solar cells. Here a controlled one-step method is

presentedwhere PbCl2 is combinedwith three equivalentsmethylammonium halide (MAX, with X¼ I and/or

Br) in polar solvents to form MAPb(I1�xBrx)3 perovskite films upon annealing in air at 145 �C. The procedure

allows for a linear increment of the semiconductor bandgap from 1.60 eV to 2.33 eV by increasing the Br

content. A transition from a tetragonal to a cubic structure is found when the Br fraction is larger than

0.3. X-ray photoelectron spectroscopy investigations shows that the increase of Br content is

accompanied by a shift of the valence band edge to lower energy. Simultaneously, the conduction band

moves to higher energy, but this shift is less pronounced. Time-resolved single-photon counting

experiments of the perovskite materials on mesoporous TiO2 show faster decay kinetics for Br

containing perovskites, indicative of improved electron injection into TiO2. Interestingly, kinetics of

MAPbI2.7Br0.3Cly on TiO2 scaffold became faster after prolonged excitation during the measurement. In

solar cell devices, MAPbI2.7Br0.3Cly yielded best performance, giving more than 14% power conversion

efficiency when used in combination with an n-type contact consisting of fluorine-doped tinoxide glass

coated with dense TiO2 and a mesoporous Al2O3 scaffold, and a p-type contact, spiro-MeOTAD/Ag.

Introduction

Hybrid (organic–inorganic) metal halide perovskite materialsare highly interesting for application in very efficient low-costsolar cells.1 Within a few years of active research of thesematerial for photovoltaic applications, power conversion effi-ciencies have already reached about 20% for small devices2 andmay soon approach and surpass the record efficiencies ofsilicon and thin lm technologies, such as Cu(In, Ga)Se2 andCdTe.

Mitzi and co-workers have studied the properties of a largerange of hybrid lead halide perovskites in detail.3 These mate-rials were used as the active layer in light emitting diodes andeld-effect transistors.3 Hybrid lead halide perovskites have alsothe potential to be used in photovoltaics, with various device

emistry-Angstrom Laboratory, Physical

eden. E-mail: gerrit.boschloo@kemi.uu.se

nd Astronomy, Molecular and Condensed

, Sweden

mistry-Angstrom Laboratory, Inorganic

eden

tion (ESI) available. See DOI:

60–21771

architectures and efficiencies reported over 20%.2,5–12 Hybridlead halide perovskite materials can be manufactured by simplesolution processes, as well as by vacuum evaporation, and theiroptical, electronic and morphological properties can be tunedby changing the chemical composition, e.g. changing theorganic cation, the metal atom and/or the halide.13–15

Tunability of the semiconductor bandgap is a very attractivefeature for solar cell materials. Specically, it allows for deviceswith a specic color, or for semi-transparent solar cells that canbe used as top cells in tandem devices. Among the hybrid leadhalide perovskites, methylammonium lead triiodide (MAPbI3)is the most investigated material, exhibiting a bandgap ofaround 1.5 eV,10,11 whereas MAPbBr3 presents a larger bandgapof 2.2 eV.11 Noh et al. were rst to demonstrate that the bandgapof these materials can be ne-tuned by mixing iodide andbromide. Interestingly, bromide-containing perovskites werefound to display better stability under moist air conditions.11 Bycomparison, MAPbCl3 possesses a large bandgap of 3.1 eV.16 Inmixed halide preparations where chloride-based precursors areused, however, no signicant inclusion of chloride is observedin the perovskite crystal structure. Instead it appears either asseparate MAPbCl3 crystals,17,18 or it disappears upon anneal-ing.19 Nevertheless, the chloride content in the precursor

This journal is © The Royal Society of Chemistry 2015

Paper Journal of Materials Chemistry A

Publ

ishe

d on

18

Sept

embe

r 20

15. D

ownl

oade

d by

Upp

sala

Uni

vers

ity o

n 09

/12/

2015

12:

40:1

1.

View Article Online

solution plays an important role in the realization of goodquality perovskite materials,18 presumably due to the formationof an intermediate phase.

In this work, we have studied the formation of MAPb(I1�x-Brx)3(Cl)y materials, where x is the fraction of bromide, usingPbCl2 as the lead precursor in a one-step solution-basedmethod. A similar approach was used recently by Suarez et al.4

We demonstrate the possibility to perform linear bandgaptuning of the hybrid lead halide perovskite, and nd an exten-sion of the stability range for the tetragonal crystalline struc-ture. A detailed characterization of the structure and the opto-electric properties of the synthetized materials are presented.Time-resolved single-photon counting experiments suggestimproved electron injection from hybrid perovskite into TiO2

when bromide is included. The kinetics found for bromidecontaining perovskites on TiO2 scaffold became faster uponprolonged excitation during the measurement. Best solar cellresults were obtained for MAPb(I1�xBrx)3(Cl)y with x ¼ 0.1 onmesoscopic Al2O3 scaffold layers.

Results and discussionSolution-based synthesis of MAPbX3 and structuralcharacterization

In this study we used PbCl2, MAI and MABr as precursors forhybrid lead halide perovskites, in combination with a mixtureof polar solvents: dimethylformamide (DMF) and dimethylsulfoxide (DMSO) in a ratio of 7 : 3. Tetrahydrofuran (THF, 10vol%) was added to the solvent mixture as it was found todecrease the conversion temperature of the organic–inorganichybrid perovskites (OIHPs) during in the annealing step. Theuse of PbCl2 as precursor for Br-containing hybrid perovskites isinspired by the use of this salt as precursor in the work bySnaith and co-workers, who obtained efficient perovskites withthe structure MAPbI3�xClx.10 In later studies, it was found thatthe chloride content of this material was less than 5%.20 In ourstudy, the ratio MAX (where X is I or Br) to PbCl2 was kept at 3 : 1in the preparation. We will refer to the formed materials asMAPb(I1�xBrx)3(Cl)y to indicate that they were made usingchloride-containing precursors. The resulting lms do,however, not contain substantial amounts of chloride, as dis-cussed elsewhere.19 The overall reaction for formation of thehybrid perovskite is proposed to be:

3MAX + PbCl2 / MAPbX3 + 2MACl,

where X is I and/or Br. Upon annealing, MACl is decomposedinto gas phase components: MACl (s) / CH3NH2 (g) + HCl (g).However, it is expected that to some degree such reactionsalready occur in the precursor solution.

Fig. 1 presents the X-ray diffractograms of MAPbI3(Cl)y andMAPbBr3(Cl)y materials spin-coated on a mesoscopic TiO2/glasssubstrate, either dried in ambient environment (25 �C) orannealed at 145 �C for 45 minutes (the substrate temperatureincreased gradually from 20 to 140 �C in this period, see ESI,Fig. S16†). During the drying process of the OIHP precursorsolution, THF, which has a boiling point of 66 �C, is expected to

This journal is © The Royal Society of Chemistry 2015

evaporate readily, while DMF and DMSO may partially remainin the dried OIHP precursor lm. This is conrmed by ATR-FTIR, which demonstrates that DMSO is still partially present inthe dried precursor lms, see Fig. 1c and S1.† The driedprecursor lms have clear XRD patterns that differ from theXRD patterns of crystalline PbCl2, PbI2, PbBr2 or MA salts (seeFig. 1(c) and S3 in ESI†). This is a clear indication that acomplexation reaction occurs between PbCl2, the halide saltsand/or solvents at low temperature. When comparing dried andannealed OIHPs, signicant differences are found betweenMAPbI3(Cl)y and MAPbBr3(Cl)y, where MAPbI3(Cl)y is trans-formed from an intermediate phase at room temperature (seeblack solid line in Fig. 1(a)) to a tetragonal perovskite structureupon annealing (space group: I4/mcm, see blue solid line inFig. S2(e)†).21 The intermediate phase of OIHP does not matchwith crystalline phases reported in literature, but is similar tothe XRD pattern reported for MACl-assisted formation OIHP.22

The MAPbBr3(Cl)y precursor dried at room temperature has acubic phase (space group: Pm3m) and is very similar to theperovskite structure obtained aer annealing at 145 �C (seeFig. 1(b)). It has a bright yellow lemon color, which changes to aslightly darker yellow color upon annealing. During thisprocess, the XRD pattern is only slightly shied to lower anglessomething that may be attributed to the vaporization of MACl(see red solid line in Fig. 1(b)).21–24

In the preparation method used, it is easy to obtain mixedMAPb(I1�xBrx)3(Cl)y perovskites, simply by adding the requiredproportion of MAI and MABr to the PbCl2-containing precursorsolution. XRD measurements (Fig. 2) show a gradual changefrom a tetragonal structure, as found for the pure MAPbI3phase, to a cubic structure, as found for the pure MAPbBr3. Thecomplete XRD patterns for the formed MAPb(I1�xBrx)3(Cl)yperovskites are shown in ESI (Fig. S2†). Similar observations formixed I/Br lead perovskite have previously been reported by Nohet al. and Eperon et al.11,25 Fig. 2(a) and (b) shows the (110)t peakthat transforms into the (100)c peak, and the (224)t peak thatbecomes (220)c. The calculated lattice parameters for orienta-tions of all OIHPs are summarized in Fig. 2(c). The crystallinityof (110)t for MAPbI3(Cl)y in this study and in our previous studyis observed very much higher than for a standard MAPbI3 lmprepared in the absence of chloride ions.21 The lattice constantschange gradually with the bromide fraction x, but a clearsingularity is found between x ¼ 0.3 and 0.4 for all crystalorientations, which we attribute to the transition from atetragonal to a cubic structure. In the study by Noh et al.11 thistransition for MAPb(I1�xBrx)3(Cl)y was found to occur betweenx ¼ 0.13 and 0.20. This suggests that the synthetic details andthe choice of precursors are crucial for the composition of thistransition, and that the presence of chloride anions in ourpreparation and extends the stability range of the tetragonalcrystalline structure. Eperon et al.25 found for FAPb(I1�xBrx)3perovskite, where FA is formamidinium, a tetragonal phase forx ¼ 0 to 0.2 and a cubic for x ¼ 0.6 to 1, while intermediatepreparations did not give crystalline phases.

The degree of crystallinity of the different lms is shown inFig. 2(d), which displays the integrated intensity of the crystal-line populations. This value is rather constant value for Br

J. Mater. Chem. A, 2015, 3, 21760–21771 | 21761

Fig. 1 XRD patterns of room temperature-dried hybrid perovskite precursors and hybrid perovskite thin films. (a) Comparison between dried andannealed OIHP samples of MAPbBr3(Cl)y. (b) Comparison of dried and annealed MAPbI3(Cl)y. (c) ATR-FTIR spectra of dried and annealedMAPbI3(Cl)y.

Journal of Materials Chemistry A Paper

Publ

ishe

d on

18

Sept

embe

r 20

15. D

ownl

oade

d by

Upp

sala

Uni

vers

ity o

n 09

/12/

2015

12:

40:1

1.

View Article Online

fractions from x ¼ 0.0 to 0.3 in MAPb(I1�xBrx)3(Cl)y. From x ¼0.4 to 1 an increase in crystallinity is found. The boundarybetween decreasing and increasing crystallinity at x ¼ 0.3–0.4lies at the same position as the transition from tetragonal tocubic phase.

The surface morphologies of the different perovskitesamples on mesoporous (ms) TiO2 were investigated usingscanning electron microscopy (SEM), see Fig. 3. In all samplespart of the underlying ms-TiO2 is visible; there is thus nocomplete over-standing layer of perovskite. The coverage wasbest (ca. 80%) for samples with x ¼ 0.3 to 0.7. Clear differencesin the perovskite morphology are observed depending on itscomposition. In MAPbI3(Cl)y many morphological defects arefound, while in the case of MAPbI2.7Br0.3(Cl)y the surfaceappears to be much denser (Fig. 3(a) and (b)). Jeon et al.reported rather similar observations of morphological differ-ences between FAPbI3 and (FAPbI3)0.85(MAPbBr3)0.15, namely,that Br improves the formation of dense, defect free perovskitelms.2

Optical absorption and emission

Thin lms of mixed MAPb(I1�xBrx)3 perovskites onms-TiO2/glasssubstrates display a gradual change of color when the fraction Bris increased, from dark brown (x ¼ 0) to yellow (x ¼ 1), see thepictures in Fig. 4. The corresponding shis in optical spectra are

21762 | J. Mater. Chem. A, 2015, 3, 21760–21771

shown in Fig. 4(a). The absorption onset of MAPb(I1�xBrx)3(Cl)ychanged systematically from 780 nm to 540 nm with increasingBr fraction, consistent with observations in recent studies.4,11,25

Notably, crystalline bulk MAPbI3 is also reported to have a higherabsorption onset of about 820 nm.22 The optical band gaps wereextracted from a Tauc plots, presented in Fig. S8 in ESI.†Based on this procedure optical bandgaps of 1.60 eV and 2.33 eVin were determined for MAPbI3(Cl)y and MAPbBr3(Cl)y, respec-tively, and the bandgaps of the intermediate compositionsMAPb(I1�xBrx)3(Cl)y vary rather linearly between these numbers(see Fig. 4(b)). The values found here are slightly larger than thosereported by Noh et al. and Suarez et al. in similar studies.4,11

An interesting observation is that the gradual increase inabsorbance at wavelengths below 650 nm for x ¼ 0.0 to 0.2, afeature which is frequently found in MAPbI3 thin lms, isabsent for x¼ 0.3 to 0.5. The origin of this absorption is not wellunderstood, but it is possibly due to the presence of amorphousOIHP, le aer incomplete reaction.26 Absence of this absor-bance can then be attributed to formation of more crystallineOIHPs for lms with Br fraction x ¼ 0.3 to 0.5 (although it isnoted that the integrated XRD intensities do not show signi-cant changes in this range (Fig. 2(d))). For x ¼ 0.6 to 1.0, theonset wavelength of the gradual increase in absorbance changesfrom 550 to 480 nm.

Photoluminescence (PL) spectra were measured on agedMAPb(I1�xBrx)3(Cl)y samples at around 30 �C for 20 days,

This journal is © The Royal Society of Chemistry 2015

Fig. 2 Transition of the tetragonal to the cubic phase in XRD diffractograms of MAPb(I1�xBrx)3(Cl)y, for x¼ 0% Br to 100% Br in 10% steps (from leftto right), shown for (a) the (110)t peak to (100)c peak and (b) (224)t to (220)c. (c) Calculated lattice constants and (d) calculated crystallinepopulation for variable ratio of Br substitution of OIHPs.

Paper Journal of Materials Chemistry A

Publ

ishe

d on

18

Sept

embe

r 20

15. D

ownl

oade

d by

Upp

sala

Uni

vers

ity o

n 09

/12/

2015

12:

40:1

1.

View Article Online

see Fig. 4(c) and S5.† Sadhanala et al. found that freshMAPb(I1�xBrx)3 lms and 21 days aged lms have signicantlydifferent PL spectra, suggesting stabilization of the structureupon aging.27 We nd a systematic change of the emission peakwith the fraction of Br, as has been demonstrated previ-ously,25,27,28 ranging from 535 nm for MAPbI3(Cl)y to 775 nm forMAPbBr3(Cl)y, see Fig. 4(e). In the range x ¼ 0.4 to 0.9 inMAPb(I1�xBrx)3(Cl)y, however, rather different PL spectra arefound: these spectra show a weak emission peak in the expectedrange between 535 and 700 nm, but a much larger peak at lowerwavelength, see Fig. 4(d) (PL spectra without normalization areshown in ESI, Fig. S5†). We attribute this to energy transfer fromthe mixed MAPb(I1�xBrx)3(Cl)y phase to a small fraction ofMAPbI3 nanocrystals, which subsequently emits. The emissionis blue shied compared to that found for the MAPbI3(Cl)ysample, which is due to either a small fraction of Br in thenanocrystals or the small size of the MAPbI3 nanocrystals, sosmall that the bandgap is increased due to quantum-size effects(quantum connement effects).29

X-ray photoelectron spectroscopy (XPS) studies

Valence band (VB) spectra of the series of MAPb(I1�xBrx)3(Cl)ymaterials deposited on ms-TiO2 are presented in Fig. 5(a). Aclose-up of the upper valence band (+3.5 to �2 eV energy range)is shown in the insert and includes a linear extrapolation and anindication of the experimental spectral edge as described in the

This journal is © The Royal Society of Chemistry 2015

Experimental section. We can observe that the valence band forthe present set of samples shi to higher binding energy versusthe Fermi level when the amount of bromine increases in thematerial. In such comparisons, it is important to remember thatthe observed Fermi level might be sensitive to subtle variationssuch as bulk doping and surface states. A difference in valenceband edge of about 0.5 eV between MAPbI3(Cl)y andMAPbBr3(Cl)y is observed.

Using the valence band data from XPS and the bandgapdetermined from optical absorption, we can construct theenergy diagrams presented in Fig. 5(b). Increase of Br content inMAPb(I1�xBrx)3(Cl)y leads to a decrease in the VB energy and anincrease in the conduction band (CB) level. The procedure usedhere indicates that the shi in VB is slightly more pronouncedthan the shi in CB.

Time-correlated single photon counting (TCSPC)

To gain further insight into the luminescence properties, wemeasured time-resolved emission of MAPb(I1�xBrx)3(Cl)ysamples. The main results are summarised in Fig. 6. Forsamples on mesoporous TiO2 with different fractions of Br (x ¼0, 0.1, 0.4, 0.7 and 1), the decay traces are given in Fig. 6(a).Clearly, the observed decays are not single exponential, but arewell-tted using a bi-exponential function. The obtained life-times and amplitudes are shown in Fig. 6(b) and (c), respec-tively. As a general trend, increasing the Br fraction leads to

J. Mater. Chem. A, 2015, 3, 21760–21771 | 21763

Fig. 3 Top view SEM images (4 mm scales, horizontal bars) of the different CH3NH3Pb(I1�xBrx)3(Cl)y perovskite deposited on a mesoscopic TiO2/glass substrate.

Journal of Materials Chemistry A Paper

Publ

ishe

d on

18

Sept

embe

r 20

15. D

ownl

oade

d by

Upp

sala

Uni

vers

ity o

n 09

/12/

2015

12:

40:1

1.

View Article Online

signicantly faster luminescence decay. A very long-livedcomponent (25% amplitude, s2 ¼ 3.5 ms) is found for perovskitewith a Br mole fraction of 0.1. The longer of the tted lifetimescomponents s2 changes from 1.0 via 3.5, 0.9, 0.5 to 0.1 ms whenincreasing x from 0 to 1, while the corresponding decreases forthe faster component s1 is from 0.18 ms to 0.01 ms. A possibleexplanation for the bi-exponential decay is that part of theperovskite material is in direct contact with mesoscopic TiO2,and shows a short lifetime due to electron injection,31 while theover-standing layer of perovskite is not able to inject directlyand gives longer PL lifetimes.32 This idea is strengthened bymeasurements of perovskite lms on glass substrates. As can beseen in Fig. 6(g) and (h) for MAPbI3(Cl)y and MAPbI2.7Br0.3(Cl)yrespectively, both perovskite samples on glass substrates, showvery long lifetimes (s2 ¼ 1.2 ms), which is indicative of high-quality photovoltaic material.33 These lms are presumablystructurally similar to the overstanding layers of perovskitelms deposited on mesoscopic scaffold layers. Surprisingly,perovskites on Al2O3 scaffold layers shows a signicant part ofrapid luminescence quenching. In this case electron injectioncannot take place. Rapid quenching must be induced by themesoscopic structure, or by interface states at the Al2O3. Aris-tidou et al.34 found a fast quenching process for MAPbI3 onmesoscopic Al2O3 that is related to the generation of superoxide

21764 | J. Mater. Chem. A, 2015, 3, 21760–21771

(O2�) on the perovskite surface formed in a photoinduced

reduction process of oxygen. Further detailed studies are,however, needed to resolve the precise origin of rapid emissionquenching in our case.

When conducting the TCSPC measurement of mixed I/Brperovskites on TiO2, the kinetics was observed to change duringthe course of the data acquisition. Early, quick measurementsshowed very long lifetimes, which became faster with increasingmeasurement times. For samples with Al2O3 and glass assubstrate, however, no such trend was observed. This is illus-trated in Fig. 6(d)–(f), where the tted lifetimes are plottedagainst the irradiation time during the measurement (given asaccumulated counts in the maximum channel). A light-inducedmodication of mixed-halide perovskites has recently beenreported by Hoke et al.35 They reported that observed uores-cence changes are accompanied by changes in the XRD patternand are thus assumed to involve structural modications andsegregation into two crystalline phases, including an iodide-rich (I-rich) one acting as quencher for the high-energy states.An important observation in our work is that light-inducedchanges were only visible on ms-TiO2 as a substrate, not onglass or ms-Al2O3 under the given illumination conditions. Thismay hint at the role of charges that are required for theformation of I-rich OIHP centers.

This journal is © The Royal Society of Chemistry 2015

Fig. 4 Top: Optical photographs of MAPb(I1�xBrx)3(Cl)y films. (a) UV-vis absorption spectra of these samples. (b) Optical bandgap as function ofBr-fraction in MAPb(I1�xBrx)3(Cl)y. (c) Photoluminescence spectra for the same films (normalized at maximum emission peak). (d) PL spectra ofMAPb(I1�xBrx)3(Cly) films scaled for bandgap emission. (e) PL emission peak energy as function of Br-fraction in MAPb(I1�xBrx)3(Cl)y.

Paper Journal of Materials Chemistry A

Publ

ishe

d on

18

Sept

embe

r 20

15. D

ownl

oade

d by

Upp

sala

Uni

vers

ity o

n 09

/12/

2015

12:

40:1

1.

View Article Online

Toluene addition to OIHP precursor was found to improvecrystallinity, coverage area and density of the lms for solar cellapplication. This effect was investigated by UV-vis spectra, XRD,

This journal is © The Royal Society of Chemistry 2015

SEM images and solar cell performances, reported in ESI,Fig. S9 to S11.† Toluene addition in OIHP precursor also gaveslightly better PL lifetimes (MAPbI3(Cl)y/Al2O3/glass: around 1.4

J. Mater. Chem. A, 2015, 3, 21760–21771 | 21765

Fig. 5 (a) Valence levels spectra of the MAPb(I1�xBrx)3(Cl)y series measured by XPS with a photon energy of 1486.6 eV. Inset: zoom of the uppervalence band (+3.5 to�2 eV energy range) including a linear extrapolation and an indication of the experimental spectral edge for theMAPbI3 andMAPbBr3 samples. (b) Schematic drawing of the MAPb(I1�xBrx)3(Cl)y energy levels determined from (a) and Fig. 4(b). The energy level scale is setversus the Fermi level using the experimental spectral edge (full arrows), i.e. 0.4 eV were removed from the linear extrapolation energies (dashedarrows).

Journal of Materials Chemistry A Paper

Publ

ishe

d on

18

Sept

embe

r 20

15. D

ownl

oade

d by

Upp

sala

Uni

vers

ity o

n 09

/12/

2015

12:

40:1

1.

View Article Online

ms (s2) and MAPbI2.7Br0.3(Cl)y/Al2O3/glass: around 0.8 ms (s2)).The tting values are shown in Fig. S7(c) and (d).†

Solar cell performance

The solar cell performance of MAPb(I1�xBrx)3(Cl)y with a lowfraction of Br (0 to 10%) was tested. In this regime, thedifferences in light harvesting efficiency due to the change inbandgap were not signicant. In Fig. 7, J–V and IPCE curves of

21766 | J. Mater. Chem. A, 2015, 3, 21760–21771

solar cells using ms-TiO2 scaffold layer are summarized.Short-circuit current densities calculated by integration ofIPCE results are in good accordance with experimental Jsc dataunder simulated sunlight. The open-circuit potential (Voc)and the short-circuit photocurrent density (Jsc) increasedfrom 0.68 to 0.82 V and from 13.7 to 21.6 mA cm�2, respec-tively, with an increase of Br content from 0 to 10%. The effectof Br content from 0 to 10% is also clearly seen in IPCE curvesin the plateau values that increase, and in the slight blue shi

This journal is © The Royal Society of Chemistry 2015

Fig. 6 Time-Correlated Single Photon Counting (TCSPC) results at 2500 counts of MAPb(I1�xBrx)3(Cl)y films. (a) TCSPC results for variation offraction of Br on OIHPs/TiO2 (x¼ 0.0, 0.1, 0.4, 0.7 and 1.0) (b) PL decay time constants s1 and s2 obtained from biexponential fits of data shown in(a). (c) Fitted amplitudes A1 and A2 from the same, (d) effect of illumination time on PL decay time constants for MAPbI3(Cl)y/TiO2 (the amplitudesare constant at A1 79% and A2 21%), (e) effect of illumination time on PL decay time constants for MAPbI2.7Br0.3(Cly)/TiO2 (the amplitudes areconstant at A1 25% and A2 75%). (f) Effect of illumination time on PL decay time constants for MAPbI2.7Br0.3(Cly)/Al2O3 (the amplitudes areconstant at A1 50% and A2 50%). TCSPC results at 20 000 counts on (g) MAPbI3(Cl)y and (h) MAPbI2.7Br0.3(Cly) with different substrates asindicated.

Paper Journal of Materials Chemistry A

Publ

ishe

d on

18

Sept

embe

r 20

15. D

ownl

oade

d by

Upp

sala

Uni

vers

ity o

n 09

/12/

2015

12:

40:1

1.

View Article Online

of the photocurrent onset edges from 800 nm to 778 nm inFig. 7(b). We believe that these enhancements are related tothe denser perovskite structure with lower amount of defectsfound for the 10% Br substituted OIHP. These results mayalso be correlated with TCSPC experiments that show fasterinitial decay for the 10% Br perovskite, which may be indic-ative of more efficient electron injection in the mesoporousTiO2, as well as a longer lived photoluminescence tail, whichmay indicate better perovskite quality in the overstandinglayer.32

The ll factors (FF) of the perovskite solar cells using ms-TiO2 scaffold layer are poor, ranging from 0.35 for MAPbI3(Cl)yto 0.48 (5% Br sample). We attribute the low FF to the poor over-standing OIHP layer that covers only part of the ms-TiO2 scaf-fold (see Fig. 3 and S10(a,b,e and f)†). This implies that there ismuch contact between the spiro-OMeTAD hole conductingmaterial (HTM) and the ms-TiO2, giving rise to signicantelectron/hole recombination.

This journal is © The Royal Society of Chemistry 2015

The use of an ms-Al2O3 scaffold layer has been reported toenhance Voc and PCE of hybrid perovskite solar cells.10

Recombination between ms-TiO2 to HTM is prevented in thisdevice structure. Using Al2O3 scaffold layer in combinationwith MAPbI3(Cl)y and MAPbI2.7Br0.3(Cl)y, we obtained muchimproved PCE values of 8.0% and 14.2%, respectively. Apossible explanation is that rapid trapping of one chargecarrier occurs in case of the presence of an Al2O3 scaffoldlayer, as may be deducted from TCSPC results. This can bebenecial charge separation.34 There is a signicant hyster-esis in the J–V curves for forward and reverse scan thesedevices. Interestingly, the composition of the perovskite had alarge effect on the degree of hysteresis. For MAPbI3(Cl)y 30%difference in the calculated PCE value was found (forward(short circuit current to open circuit voltage): 5.6% andbackward (open circuit voltage to short circuit current): 8.0%),while for MAPbI2.7Br0.3(Cl)y it was less than 10% (PCEforward: 14.2% and backward: 12.8%). This different J–V

J. Mater. Chem. A, 2015, 3, 21760–21771 | 21767

Fig. 7 Solar cell performances for different ratio of Br substituted OIHPs; (a) J–V curve (TiO2 scaffold layer), (b) IPCE curves (TiO2 scaffold layer),(c) J–V curves of MAPbI3(Cly) and MAPbI2.7Br0.3(Cly) (Al2O3 scaffold layer), (d) summary of PCE values obtained for perovskite solar cells with TiO2

or Al2O3 scaffold layer with different fractions of Br.

Journal of Materials Chemistry A Paper

Publ

ishe

d on

18

Sept

embe

r 20

15. D

ownl

oade

d by

Upp

sala

Uni

vers

ity o

n 09

/12/

2015

12:

40:1

1.

View Article Online

hysteresis can be attributed to the improved quality of OIHPlm that contains a fraction of Br.

Conclusion

A convenient 1-step synthesis method for the mixed hybridperovskite MAPb(I1�xBrx)3(Cl)y, using PbCl2 as lead source, wasinvestigated. The boundary between tetragonal and cubic crys-talline phase was found to occur between x ¼ 0.3 and 0.4. Theoptical bandgap varied linearly with the bromide fraction xfrom 1.60 to 2.33 eV. PL spectra showed the same gradualchange, however, at intermediate x-values additional andstronger emission was found at lower energy, attributed toenergy transfer of a small fraction of I-rich OIHP quantum dots.

Time-resolved single-photon counting investigations suggestthat inclusion of bromide leads to improved electron injectionfrom hybrid perovskite into mesoporous TiO2. Prolonged exci-tation during this measurement led to faster decay kinetics ofbromide containing perovskites on TiO2 scaffold, while theywere unchanged on mesoscopic Al2O3. Best solar cell results

21768 | J. Mater. Chem. A, 2015, 3, 21760–21771

were obtained for MAPb(I1�xBrx)3(Cl)y with x ¼ 0.1 on meso-scopic Al2O3 scaffold layers, yielding 14% efficiency.

MethodsMaterials and sample fabrication for characterization

MAI was synthesized by reacting 360 mL methylamine (40% inwater, Sigma Aldrich), 280 mL methylamine (40% in methanol,TCI) and 40 mL of hydroiodic acid (without stabilizer, distilled57 wt% in water, Aldrich) in a 1000 mL round bottom ask ataround 0 �C for 5–7 h with stirring. The precipitate was recoveredby evaporation at 70 �C for 3 h. The MAI was washed with diethylether by stirring the solution for 30 min, which was repeatedthree times then nally dried at 70 �C in a vacuum oven for 24 h.The yellowish colored MAI (MW 158.96) product was obtained.MABr (MW 111.97) was synthesized as described elsewhere.11

Precursor solutions MAPbI3(Cl)y, MAPb(I1�xBrx)3(Cl)y andMAPbBr3(Cl)y were formed by mixing a ratio of three to one ofMAI (0.9 to 0.0 M) and MABr (0.0 to 0.9 M) with PbCl2 (0.3 M) indimethylformamide (DMF)/dimethyl sulfoxide (DMSO) (7/3 ¼v/v). PbCl2 was rst completely dissolved by stirring on a hot-plate

This journal is © The Royal Society of Chemistry 2015

Paper Journal of Materials Chemistry A

Publ

ishe

d on

18

Sept

embe

r 20

15. D

ownl

oade

d by

Upp

sala

Uni

vers

ity o

n 09

/12/

2015

12:

40:1

1.

View Article Online

set at 70–80 �C for few hours, aer then they reacted with MAI orMABr (mixed MAX: X ¼ Br and I) for 2 hours. Finally, tetrahy-drofuran (THF, 10 vol%) was added. TiO2 nanoparticles weredeposited on microscope glass by spin coating method at 3000rpm for 30 s. The TiO2 suspension was Dyesol paste (DSL 30 NR-D) diluted with terpinol 25 wt% in ethanol (1 : 4 ¼ g/g). The TiO2

nanoparticle coated microscope glass was annealed onto hotplate at 550 �C for 30 min. The lm thickness was approximately300 nm (�50 nm) as determined by a Dektak3 proler. Unlessstated otherwise, OIHP precursors are coated onto TiO2/glasssubstrate by spin-coating method at 1500 rpm for 30 s. These areannealed onto hot plate at 145 �C for 30–45 min in a dry box(humidity & 20%), see Fig. S16.†

ATR FT-IR spectra

ATR FT-IR spectra were obtained from the powders of MACl,MABr, MAI, dried MAPbBr3(Cl)y and MAPbI3(Cl)y which werecarried on a Nicolet Avatar 370 DTGS spectrophotometer(Thermo Co. LTD., resolution is 0.964 cm�1, scan counting has300 times per a sample). Background signal was subtracted.

UV-vis-NIR spectra

To measure UV-visible-NIR absorption spectra, organic–inor-ganic perovskite coated TiO2 electrodes deposited on workingelectrode were prepared according to the mention above andthe measurements were carried on a Varian Cary 5000 UV-vis-NIR spectrophotometer. The FTO glass substrate signal wasused as calibration background.

Steady-state emission measurements

Standard steady-state emission spectra were obtained on aFluorolog-3 instrument (Horiba Jobin Yvon) equipped withdouble-grating excitation and emission monochromators and a450 W Xe lamp as a light source. The emission spectra werecorrected for the spectral sensitivity of the detection system byusing a calibration le of the detector response. Front-faceillumination (30� with respect to the incident beam) was used tominimize inner-lter effects. The pump laser had setup atrelevant wavelength to the valence band of OIHPs. The ltershas been setup front of detectors.

Time-correlated single photon counting (TCSPC)

A detailed description of the experimental setup has been givenrecently.36 Briey, the sample was excited on the perovskite sidewith 77.1 ps pulses of 404.6 nm from a picosecond diode laser(Edinburgh Instruments, EPL-405). Certain measurements weredone with 470 nm excitation (EPL-470). As practically the sameresults were obtained, we used 404.6 nm excitation because ofthe shorter instrument response function due to shorter exci-tation pulses. The laser pulse energy was very low, ca. 15 pJ andwas attenuated even further to reach the desired count rate ofca. 1%. This ensures low enough excitation to avoid Augerrecombination37 in OIHPs and that the results are free frompulse pile-up.38,39 Measurements where done in reverse mode at5–10 MHz and under magic angle polarization. A cutoff lter

This journal is © The Royal Society of Chemistry 2015

(OG590) was used to block stray excitation light. A dilute solu-tion of Ludox was used to record the instrument responsefunction without any lter. No monochromator was used, i.e. allwavelengths transmitted by the cutoff lter were collected,increasing the instrumental sensitivity.

Crystallographic properties

The OIHP deposited on TiO2/glass substrate were investigatedfor crystallographic properties using X-ray diffraction (XRD)with a Siemens diffraktometer D5000 apparatus at roomtemperature using CuKa radiation in time period. Diffrac plusXRD commander program was used to control the instrument.The instrument was set in “detector scan” mode, and theacquisition was done in q–2q mode for every 0.02� incrementover the Bragg angle range of 10–60�.

Scanning electron microscopy (SEM)

Scanning Electron Microscopy (SEM) was performed on a Zeiss(Gemini 1550) microscope having a eld emission (FE) electronsource and an in-lens detector for secondary electrons. Top viewimages were recorded using a high tension of 10 kV.

X-ray photoelectron spectroscopy (XPS)

XPS measurements were performed with a Scienta ESCA 300instrument, using monochromatized AlKa radiation hn ¼1486.7 eV with an overall instrumental resolution of 0.6 eV. Thepressure in the analysis chamber was around 1 � 10�8 mbarand the electron take off angle was 90�. The valence spectrapresented were binding energy calibrated using the Ti 2p corelevel signal from the TiO2 substrate which in this study was setto 459.45 eV. This value was obtained by measuring the energydifference between the Ti 2p core level of pure TiO2 and theFermi level on an Au sample with the same instrument. Thiscalibration was crosschecked with the position of the metalliclead sometimes observed and found at 137.0 eV versus the Fermilevel of the Au sample. Charging and radiation effects werechecked for by measuring the specic core level repetitively andwas found negligible for all spectra reported in the presentinvestigation. The shape of the VB will depend on the photonenergy used.40 To evaluate the relative changes in the upperenergies of the valence band, we used a linear extrapolation.The linear extrapolation energy is dened as the intersection of alinear extrapolation of outermost experimental valence levelswith the baseline. Moreover, in the spectra of MAPbI3 andMAPbBr3 we indicate an experimental spectral edge were theamount of counts above the baseline is less than about onepercent of the maximum value at about 4.0 eV. For both MAPbI3and MAPbBr3 we note that the difference between the experi-mental spectral edge and linear extrapolation energy is close to 0.4eV. In the present paper, we therefore remove 0.4 eV to thelinear extrapolation energy as an estimation of the valence bandedge when compared to the Fermi level.30 This procedure givesvalues that relate to literature values of the difference between aFermi level of TiO2 pinned to the conduction band edge and ananatase bandgap of 3.2 eV.

J. Mater. Chem. A, 2015, 3, 21760–21771 | 21769

Journal of Materials Chemistry A Paper

Publ

ishe

d on

18

Sept

embe

r 20

15. D

ownl

oade

d by

Upp

sala

Uni

vers

ity o

n 09

/12/

2015

12:

40:1

1.

View Article Online

Fabrication and measurement of solar cell devices

Fabrication of solar cell device. All experiments are carriedout in 30%–45% humidity. The OIHP precursors were preparedas follow as in the section of material and sample fabrication forcharacterization. MAPbI3Cly, MAPbI3�xBrxCly and MAPbBr3Clywere formed by mixing a ratio of three to one of MAI (3.60 M(100%)–3.24 M (90%)) and MABr (0.00 M (0%)–0.36 M (10%))with PbCl2 (1.2 M) in DMF/DMSO (7/3 ¼ v/v) on a hot plate of70–80 �C for 90 min. For the solar cell devices, tetrahydrofuran(THF)/toluene (1/2 ¼ v/v) mixed solvent was added to the OIHPprecursor and it is consisted of 85% precursor and 15% THF/toluene solvent (v/v). The effect of toluene on morphology ofOIHP has well been investigated in previous report.12 All OIHPprecursors are kept on 70 �C on a hot plate during fabricationprocesses. TiO2 precursor was prepared as follow as in thesection of Materials and sample fabrication for characteriza-tion. Al2O3 precursor purchased from Sigma Aldrich which issize of �50 nm in isopropanol and diluted with isopropanol as1 : 2 (v/v). Fluorine-doped tin oxide (FTO) coated glass(Pilkington TEC 15) 15 U ,�1 was patterned according toetching processes with Zn powder and 2MHCl diluted in water.TiO2 compact layer as electron extracting layer was depositedthe lm thickness of 40–70 nm onto the eached FTO substrateand temperature of hot plate was kepted at 400–450 �C duringspray pyrolysis deposition process. The prepared TiO2 and Al2O3

precursors were spincoated for 20 s at 4000 and 2750 rpm,respectively, and dried on a hot plate at 450 �C (TiO2) or 150 �C(Al2O3) for 30 min. The prepared OIHP precursor was spincoated at 1250 rpm for 20 s and annealed at 145 �C for 30–45min. The hole transporter was deposited by spin-coating at 1250rpm for 20 s using an 8 wt% 2,20,7,70-tetrakis-(N,N-di-p-methoxyphenyl-amine)9,9-spirobiuorene (spiro-OMeTAD) inchlorobenzene with added tert-butylpyridine (TBP) and lithiumbis(tri-uoromethanesulfonyl)imide (Li-TFSI) of 80 and 30mol%, with respect to spiro-OMeTAD. The nal producingspiro-OMeTAD was oxygen doped.41 Finally, 150 nm thick silverelectrodes were deposited on top of devices by thermal evapo-ration at �10�6 bar, through a shadow mask.

Power conversion efficiency (PCE). The light source of a solarsimulator for measuring the current–voltage (J–V) characteris-tics was a 300 W solar simulator (Newport) calibrated to a 1000W m�2 light intensity at the 1.5 AM Global condition (1 sun AM1.5 G illumination) by a certied silicon solar cell (FraunhoferISE). The electrical data were recorded with a computercontrolled digital source-meter (Keithley Model 2401) with thescan direction from the open-circuit to the short-circuit at ascan rate of 800–1250 mV s�1.42,43 TiO2 scaffold based solar cellswere masked during the measurement with an metal mask witha square aperture of 0.2 cm2 (0.10 cm2 for Al2O3 scaffold basedsolar cells).

Incident photon to current conversion efficiency (IPCE). TheIPCE spectra were recorded with a computer-controlled setupcomprised of a xenon lamp (Spectral Products, ASB-XE-175), amonochromator (Spectral Products, CM110), chopped at 30 Hz,a potentiostat (PINE instrument Company, Model AFRDE5) anda lock-in amplier (SRS830), using white LED bias light. The

21770 | J. Mater. Chem. A, 2015, 3, 21760–21771

setup was calibrated with a certied silicon solar cell(Fraunhofer ISE) prior to the measurements. TiO2 scaffoldbased solar cells were illuminated from the WE side using ablack mask with an rectangular aperture of 0.20 cm2 (0.10 cm2

for Al2O3 scaffold based solar cells). The white LED bias lightwas approx. 0.08 sun in intensity.

Acknowledgements

We thank the Swedish Energy Agency, the STandUP for Energyprogram, the Swedish Research Council (VR), the GoranGustafsson Foundation, and the Knut and Alice WallenbergFoundation for nancial support. B-w. P. thanks the membersof the Korean Scandinavian Scientist Engineers Association(KSSEA) and chairman Dr Keunjae Kim for support and regards,in particular Taeja Kim-Bjorklund, former chairman in KSSEAof Sweden. Further thanks to Mr Seockjeong Eom, Ambassadorin Embassy of Republic of Korea in Sweden.

References

1 M. A. Green, A. Ho-Baillie and H. J. Snaith, Nat. Photonics,2014, 8, 506–514.

2 N. J. Jeon, J. H. Noh, W. S. Yang, Y. C. Kim, S. Ryu, J. Seo andS. I. Seok, Nature, 2015, 517, 476–480 Efficiency chart of NREL,http://www.nrel.gov/ncpv/images/efficiency_chart.jpg.

3 D. B. Mitzi, S. Wang, C. A. Feild, C. A. Chess and A. M. Guloy,Science, 1995, 267, 1473–1476.

4 B. Suarez, V. Gonzalez-Pedro, T. S. Ripolles, R. S. Sanchez,L. Otero and I. Mora-Sero, J. Phys. Chem. Lett., 2014, 5(10),1628.

5 B. Cai, Y. Xing, Z. Yang, W.-H. Zhang and J. Qiu, EnergyEnviron. Sci., 2013, 6, 1480–1485.

6 J. H. Heo, D. H. Song and S. H. Im, Adv. Mater., 2014, 26,8179–8183.

7 Y. Zhao and K. Zhu, J. Am. Chem. Soc., 2014, 136(35), 12241–12244.

8 A. Kojima, K. Teshima, Y. Shirai and T. Miyasaka, J. Am.Chem. Soc., 2009, 131(17), 6050.

9 J.-H. Im, C.-R. Lee, J.-W. Lee, S.-W. Park and N.-G. Park,Nanoscale, 2011, 3(10), 4088–4093.

10 M. M. Lee, J. Teuscher, T. Miyasaka, T. N. Murakami andH. J. Snaith, Science, 2012, 338(6107), 643–647.

11 J. H. Noh, S. H. Im, J. H. Heo, T. N. Mandal and S. I. Seok,Nano Lett., 2013, 13(4), 1764–1769.

12 N. J. Jeon, J. H. Noh, Y. C. Kim, W. S. Yang, S. Ryu andS. I. Seok, Nat. Mater., 2014, 13, 897–903.

13 A. Kojima, M. Ikegami, K. Teshima and T. Miyasaka, Chem.Lett., 2012, 41, 397–399.

14 J. L. Knutson, J. D. Martin and D. B. Mitzi, Inorg. Chem.,2005, 44, 4699–4705.

15 K. D. Karlin and D. B. Mitzi, in Prog. Inorg. Chem., JohnWiley& Sons, Inc., New York, 2007, pp. 1–121.

16 B. Philippe, B.-W. Park, R. Lindblad, J. Oscarsson,S. Ahmadi, E. M. J. Johansson and H. Rensmo, Chem.Mater., 2015, 27(5), 1720–1731.

This journal is © The Royal Society of Chemistry 2015

Paper Journal of Materials Chemistry A

Publ

ishe

d on

18

Sept

embe

r 20

15. D

ownl

oade

d by

Upp

sala

Uni

vers

ity o

n 09

/12/

2015

12:

40:1

1.

View Article Online

17 B.-w. Park, B. Philippe, T. Gustafsson, K. Sveinbjornsson,A. Hagfeldt, E. M. J. Johansson and G. Boschloo, Chem.Mater., 2014, 26(15), 4466–4471.

18 B.-w. Park, S. M. Jain, X. Zhang, A. Hagfeldt, G. Boschloo andT. Edvinsson, ACS Nano, 2015, 9(2), 2088–2101.

19 E. L. Unger, A. R. Bowring, C. J. Tassone, V. L. Pool, A. G. -Parker, R. Cheacharoen, K. H. Stone, E. T. Hoke,M. F. Toney and M. D. McGehee, Chem. Mater., 2014,26(24), 7158–7165.

20 J. You, Z. Hong, Y. Yang, Q. Chen, M. Cai, T.-B. Song,C.-C. Chen, S. Lu, Y. Liu, H. Zhou and Y. Yang, ACS Nano,2014, 8(2), 1674–1680.

21 T. Baikie, Y. Fang, J. M. Kadro, M. Schreyer, F. Wei,S. G. Mhaisalkar, M. Graetzel and T. J. White, J. Mater.Chem. A, 2013, 1, 5628–5641.

22 Y. Zhao and K. Zhu, J. Phys. Chem. C, 2014, 118(18), 9412–9418.

23 I. P. Swainson, R. P. Hammond, C. Soulliere, O. Knop andW. Massa, Solid State Chem., 2003, 176, 97–104.

24 R. D. Waldron, J. Chem. Phys., 1953, 21, 734.25 G. E. Eperon, S. D. Stranks, C. Menelaou, M. B. Johnston,

L. M. Herz and H. J. Snaith, Energy Environ. Sci., 2014, 7,982–988.

26 J. J. Choi, X. Yang, Z. M. Norman, S. J. L. Billinge andJ. S. Owen, Nano Lett., 2014, 14(1), 127–133.

27 A. Sadhanala, F. Deschler, T. H. Thomas, S. E. Dutton,K. C. Goedel, F. C. Hanusch, M. L. Lai, U. Steiner, T. Bein,P. Docampo, D. Cahen and R. H. Friend, J. Phys. Chem.Lett., 2014, 5, 2501–2505.

28 G. Xing, N. Mathews, S. S. Lim, N. Yantara, X. Liu, D. Sabba,M. Gratzel, S. Mhaisalkar and T. C. Sum, Nat. Mater., 2014,13, 476–480.

29 G. V. Prakash, K. Pradeesh, R. Ratnani, K. Saraswat,M. E. Light and J. J. Baumberg, J. Phys. D: Appl. Phys., 2009,42, 185405.

This journal is © The Royal Society of Chemistry 2015

30 T. Leijtens, S. D. Stranks, G. E. Eperon, R. Lindblad,E. M. J. Johansson, I. J. McPherson, H. Rensmo, J. M. Ball,M. M. Lee and H. J. Snaith, ACS Nano, 2014, 8(7), 7147–7155.

31 K. G. Stamplecoskie, J. S. Manser and P. V. Kamat, EnergyEnviron. Sci., 2015, 8, 208–215.

32 M. De Bastiani, V. D'Innocenzo, S. D. Stranks, H. J. Snaithand A. Petrozza, APL Mater., 2014, 2, 081509.

33 S. D. Stranks, G. E. Eperon, G. Grancini, C. Menelaou,M. J. P. Alcocer, T. Leijtens, L. M. Herz, A. Petrozza andH. J. Snaith, Science, 2013, 342, 341.

34 N. Aristidou, I. Sanchez-Molina, T. Chotchuangchutchaval,M. Brown, L. Martinez, T. Rath and S. A. Haque, Angew.Chem., Int. Ed., 2015, 54, 8208–8212.

35 E. T. Hoke, D. J. Slotcavage, E. R. Dohner, A. R. Bowring,H. I. Karunadasa and D. McGehee, Chem. Sci., 2015, 6,613–617.

36 A. E. -Zohry, A. Orthaber and B. Zietz, J. Phys. Chem. C, 2012,116, 26144–26153.

37 G. Xing, N. Mathews, S. S. Lim, N. Yantara, X. Liu, D. Sabba,M. Gratzel, S. Mhaisalkar and T. C. Sum, Nat. Mater., 2014,13, 476–480.

38 D. V. O'Connor and D. Phillips, Time-Correlated Single PhotonCounting, Academic Press, London, 1984.

39 J. R. Lakowicz, Principles of Fluorescence Spectroscopy,Springer, Berlin, 3rd edn, 2006.

40 S. J. H., J. Electron Spectrosc. Relat. Phenom., 1976, 8, 129–137.41 U. B. Cappel, T. Daeneke and U. Bach,Nano Lett., 2012, 12(9),

4925–4931.42 R. S. Sanchez, V. G. Pedro, J.-W. Lee, N.-G. Park, Y. S. Kang,

I. Mora-Sero and J. Bisquert, J. Phys. Chem. Lett., 2014, 5,2357–2363.

43 E. L. Unger, E. T. Hoke, C. D. Bailie, W. H. Nguyen,A. R. Bowring, T. Heumuller, M. G. Christoforod andM. D. McGehee, Energy Environ. Sci., 2014, 7, 3690–3698.

J. Mater. Chem. A, 2015, 3, 21760–21771 | 21771